Catholijn Jonker*, Vagan Terziyan**, Jan Treur* * AI Department, Vrije Universitejt Amsterdam

1. Temporal and Spatial Analysis to Personalise an Agent's Dynamic Belief, Desire, and Intention Profiles Catholijn Jonker*, Vagan Terziyan**, Jan Treur* * AI Department, Vrije Universitejt Amsterdam ** MIT Department, University of Jyväskylä CIA – 2003 Helsinki, August 29, 2003

2. Authors • Catholijn Jonker • Department of Artificial Intelligence • Vrije Universiteit Amsterdam (the Netherlands) • jonker@cs.vu.nl • http://www.cs.vu.nl/~jonker • Vagan Terziyan • Department of Mathematical Information Technology • University of Jyväskylä (Finland) • vagan@it.jyu.fi • http://www.cs.jyu.fi/ai/vagan • Jan Treur • Department of Artificial Intelligence • Vrije Universiteit Amsterdam (the Netherlands) • treur@cs.vu.nl • http://www.cs.vu.nl/~treur

3. Agent is Part of Intentional System • An agent is considered as a part of intentional system and thus it is an entity which appears to be the subject of beliefs, desires, intentions, etc; • An agent is assumed to decide to act and communicate based on its beliefs about its environment and its desires and intentions. These decisions, and the intentional notions by which they can be explained and predicted, generally depend on circumstances in the environment, and, in particular, on the information on where and when these circumstances acquired.

4. Where We Might Need Spatial Considerations in an Intentional System ? • … in many applications where agent’s intentional profile essentially depends on its location in the environment, e.g.: • for adaptive location-based “Push” services for mobile customers; • for intelligent tracking of terrorists; • … • etc.

5. Adaptive Location-Based “Push” Services for Mobile Customers • It will be very helpful to have capabilities to predict in which places of the environment certain desires or intentions of a customer are likely to arise, to stimulate the arising of these intentions by providing the occurrence of circumstances that are likely to lead to them.

6. Intelligent Tracking of e.g. Terrorists, etc. • Also we assume that it may be very helpful to have capabilities to predict in which places of the environment certain inappropriate desires or intentions are likely to arise, either: • to avoid the arising of these intentions by preventing the occurrence of circumstances that are likely to lead to them, or • if these circumstances cannot be avoided, by anticipating consequences of the intentions.

7. Basic Ontologies Used • Actual state of the external world: EWOnt; • Observation observation_result(p) and communication (communicated_by(p, C) ) input of an agent: InOnt ; • Output of an agent (decisions to do actions): OutOnt ; • Agent internal ontology: IntOnt ; • Agent interface ontology:InterfaceOnt =InOntOutOnt ; • Agent ontology: AgOnt =InterfaceOntIntOnt; • Overall ontology: OvOnt =EWOntAgOnt . p – a state property of the external world; C – an agent who provides information about state property.

8. Overall Trace • The set of possible states of the overall ontology: IS(OvOnt). • An overall traceM over a time frame T is a sequence of states over the overall ontology OvOntover time frame T:(Mt)t Tin IS(OvOnt) . • States of agent A input / output interfaces and internal state at time t, given an overall trace M: • state(M, t, input(A)); • state(M, t, output(A)); • state(M, t, internal(A)).

9. Temporal Belief Statement • An agent believes a fact if and only if it received input about it in the past and the fact is not contradicted by later input of the opposite. MWt1 [ (M , t1)t0 ≤ t1 [ Input(, t0, M ) t  [t0, t1] Input(~, t, M ) ] ] Temporal belief statement

10. Desires • Given a desire, for each relevant action there is an additional reason, so that if both the desire is present and the agent believes the additional reason, then the intention to perform the action will be generated. • Every intention is based on a desire, i.e., no intention occurs without desire

11. Intentions and Actions(temporal vs. spatial) • Under appropriate circumstances an intention leads to an action: • an agent who intends to perform an action will execute the action immediately when an opportunity occurs; • an agent who intends to perform an action will execute the action at the nearest known place where an opportunity occurs.

12. Satisfaction Relation •  is true in this state at time t: state(M, t, input(A)) |=  , where SPROP(InOnt) state properties based on ontology

13. Spatial Language Elements • Location (x, y) has property p:location_has_property(x, y, p) ; • Agent A is present at location (x, y):location_has_property(x, y, present (A)) ; • For example: state(M, t, input(A)) |=observation_result(location_has_property(x, y, p)) • means that at time t the agent A's input has information that it observed that location (x, y) has property p.

14. A Route • A routeR is defined as a mapping from distances d (on the route) to locations (x, y), • e.g., after 300 m on the route home_from_school you are at location (E, 5) on the map. at_ distance_at_location (R, d, x, y) at_ distance_at_location (home_from_school, 300, E, 5)

15. Associated Route of “Walking” Agent • A trace M of an agent walking in a city specifies an associated route R(M) as follows: at route after distance d you are at location (x, y) if and only if a time point t exists such that at t agent A has walked d from the start and is present at location (x, y). at_ distance_at_location (R(M), d, x, y)   t [ state(M, t, EW) |= distance_from_start(d)   location_has_property(x, y, present(A)) ]

16. Temporal vs. Spatial Factors (Case 1) • Person A (1981-1986 M.Sc. studies on Applied Mathematics; 1987-2000 – Ph.D. studies on Artificial Intelligence; 2001-2002 – Project Work on Ontology Engineering); • Person B (M.Sc. studies on Applied Mathematics in University of Jyvaskyla; Ph.D. studies on Artificial Intelligence in Massachusetts Technological Institute; Project Work on Ontology Engineering in Vrije Universiteit Amsterdam). • spatial history here (i.e. second description) seems to be more informative than temporal history in a reasonable context.

17. Temporal vs. Spatial Factors (Case 2) • Person A (10:00 wants a cup of coffee; 15:00 wants to eat; 19:00 wants to watch TV News; 23:00 wants to sleep); • Person B (wants a cup of coffee in train “Jyvaskyla-Helsinki” near Pasila station; wants to eat in Helsinki University Conference Room; wants to watch TV News in the Irish Pub in Downtown Helsinki; wants to sleep in “Scandic” Hotel). • alternatively temporal history here (i.e. first description) seems to provide more information about a person than the spatial one in a reasonable context.

18. Predicting Agent’s States based on Spatial History Given - set of routes M for the agent with observed agent state in different location points of each route; task – online prediction of agents next locations, BDI attributes and states for a new route.

19. Relationships between BDI Notions

20. Mobile Commerce (Location-Based Service) • Agent – mobile customer. • Agent’s location – can be tracked by positioning infrastructure. • Observable agents actions – e.g. clickstream (points of interest) on a map delivered to the mobile terminal, calls and downloads of information about points of interest, appropriate orders, reservations, payments, etc. - can be tracked by Location-Based Service (LBS).

21. Mobile Location-Based Service (advanced personalization)

22. Spatial BDI Axiom • If a customer has absolutely same beliefs about content and quality of two different services, and such content and quality fits his recent desires, then this customer intentsto select nearest one to get service from it.

23. Spatial Belief Axiom • Spatial Belief Axiom: customer believes QoS(q), i.e., that in the same location he can get likely the same quality of service q as he used to get in this location before (observation and communication results) MWt1 [location_has_property(x, y,QoS(q))(M , t1)  t0 ≤ t1 [ Input(location_has_property(x, y, QoS(q)), t0, M)  t  [t0, t1] Input(~location_has_property(x, y, QoS(q)), t, M)]]

24. Tracking Spatial Beliefs • Tracking spatial beliefs is based on the Spatial Belief Axiom: the customer’s beliefs can be tracked based on his current location coordinates by analyzing the history of his observations and actions (e.g., orders) in this or neighbour locations

25. Agent’s Spatial Belief Example

26. Spatial Desire Axiom • Spatial Desire Axiom: customer being elsewheredesires to get service qin some location x,y, i.e., he believes QoS(q), “turns his view” to location x,y and also believes that there is no such service q1 with better quality closer on his route to x,y. MWt, x, y, d, q { at_distance_at_location(R (M), d, x, y)  [ get_service(x, y,q)(M , t)   t0 ≤ t  t1[t0, t] [ ( location_has_property(x, y,QoS(q))(M , t1) )   x1,y1, d1 [ ( at_distance_at_location(R (M), d1, x1, y1)  (d > d1) )  q1 ( q1 < q location_has_property(x1, y1,QoS(q1))(M , t1) ) ]  t2[t0, t] ( click_location(x, y)(M , t2) ) ] }, whereclick_location(x, y)(M , t2)denotes customers action «click to point x,y on his terminal screen», i.e. «turning view» to that point.

27. Tracking Spatial Desires • Tracking spatial desires is based on Spatial Desire Axiom: the customer’s desires can be tracked based on types and coordinates of points of interest he clicks on the screen of mobile terminal

28. Agent’s Spatial Desire Example

29. Spatial Intention Axiom • Spatial Intentions Axiom: customer intends to get some service at a certain location, i.e. he desires to get this service, and he either order/reserve this service online* or moves towards this service location point**. MWt, x, y, d, q {at_distance_at_location(R (M), d, x, y)   [ get_service(x, y,q)(M , t) ] t0 ≤ t  t1[t0, t] [(get_service(x, y,q)(M, t1))  d1,d2 at_distance_at_location(R (M), d1, x, y)   [ t2[t1, t] [ (at_distance_at_location(R (M), d2, x, y) (d1 > d2 > d) )   t3[t1, t] ( order_service_online (q)( M , t3) ) ] ] ] }

30. Tracking Spatial Intentions • Tracking spatial intentions: is based on Spatial Intentions Axiom: the customer’s intentions can be tracked either based on the evidence of his ordering/reserving the desired service online or based on sequentially decreasing distance between customer location and desired service location

31. Agent’s Spatial Intentions Examples * **

32. Spatial Action Axiom • Spatial Actions Axiom: a customer performs an action of getting some service, i.e. this service was intended, customer reaches the service location point and spends there at least minimal estimated time for this type of service or makes online electronic payment for it. MWt, x, y, d, q { at_distance_at_location(R (M), d, x, y)   [get_service(x, y,q)(M , t) ] t0 ≤ t  t1[t0, t] [(get_service(x, y,q)(M, t1))  at_distance_at_location(R (M), d1, x, y)   [ ti [t2, t3]  [t1, t] [ ( at_distance_at_location(R (M), 0, x, y)   ( t3 - t2 > min_time(q)) )  t3[t1, t] ( pay_for_service_online (q)( M , t3) ) ] ] ] }

33. Tracking Spatial Actions • Tracking spatial actions is based on Spatial Actions Axiom, customer’s actions can be tracked either based on the evidence of his electronic payment for the intended service online or based on a fact that customers coordinates are the same as intended service coordinates during minimal estimated time for this type of service.

34. Agent’s Spatial Action Example

35. Tool for Mobile Customers vs. Tool for Agents • Tool for Mobile Customers: Location-Based Service Navigator (LBSN) - helps its customers (mobile terminal users): (a) to navigate within unknown geographical locations, (b) to access information resources of real world services located in neighbourhood areas. • Tool for Agents: Autonomous Sensor Support for Agents (ASSA) - provides an “autonomous sensor” to an agent for observing environment around his location, navigating within this environment, finding, communicating with, getting knowledge about, and services from other agents (services). By keeping records of all transactions, ASSA is able to create really powerful collection of data about agents’ behaviour. Appropriate data mining and knowledge discovery algorithms can be applied to discover useful patterns of each agent spatio-temporal behaviour and use these patterns for online prediction of agents’ preferences, beliefs, desires, intentions and actions.

36. ASSA Benefits from LBSN (Changing Default Preferences) • Changing Default Preferences: • LBNS: Input: built-in default preferences screen form, which consists of scale slices and types of services to be shown for each scale slice; Action: customer edits and saves form; Output: updated default customer preferences; Comment: for example at a Street Network scale slice customer wants to see museums, hotels and restaurants only, at a City Network scale slice he prefers to see only gasoline stations, at a Country Network scale he prefers to see only embassies. • ASSA gets explicitly agent’s preferences, which can be treated as a set of possible agent’s desires.

37. ASSA Benefits from LBSN (Locating) • Locating: • LBNS: Input: customer’s request ; Action: LBSN contacts positioning infrastructure, requests to locate customer, gets coordinates and delivers to customer; Output: customer’s coordinates, four natural numbers (latitude, longitude, attitude, time point) according to agreed standard. • ASSA gets explicitly agent’s location in different time points, i.e. can collect agent’s routes within the environment and make grounded guesses based on this data about agent’s spatial BDI

38. ASSA Benefits from LBSN (Showing Location) • Showing Location: • LBNS: Input: customer’s coordinates, scale of map visualizing; Action: LBSN selects appropriate geographical data, prepares it and delivers to customer’s terminal; Output: screen with scaled map and pointed customer location on it; Comment: customer location supposes to be in the middle of screen, i.e. customer gets view of some scaled radius around his location. • ASSA gets explicitly the “picture of what the agent can see now”, i.e. can discover some peace of the agent’s knowledge .

39. ASSA Benefits from LBSN (Showing Services) • Showing Services: • LBNS: Input: customer’s coordinates, scale of map visualizing; Action: LBSN selects appropriate geographical and service data, prepares it and delivers to customer’s terminal; Output: screen with scaled map, pointed customer location on it and services, i.e. points of interest; Comment: all shown services are preliminary classified to types, displayed e.g. with different colours or different geometrical primitives for different type of service, and filtered against default user preferences. • ASSA gets explicitly data about “which services the agent can observe now”, i.e. can discover some peace of the agent’s knowledge.

40. ASSA Benefits from LBSN (Zooming) • Zooming: • LBNS: Input: screen with scaled map and pointed customer location on it, new scale of map visualizing; Action: LBSN updates map according to the new scale, prepares it and delivers to customer’s terminal; Output: screen with map updated according to a new scale and pointed customer location on it. • ASSA gets explicitly data about changing possible agents desires from one set of preferences to another one based on changes in scale. ASSA also gets new agent’s view to the neighbouring environment.

41. ASSA Benefits from LBSN (Intelligent Zooming) • Intelligent Zooming: • LBNS: Input: screen with scaled map, pointed customer location and services on it, new scale of map visualizing, preferences filter associated with new scale; Action: LBSN updates and filters appropriate service data according to the new scale, prepares it and delivers to customer’s terminal; Output: screen with map and services updated according to a new scale and preference filter and pointed customer location on it; Comment: independently of selected scale LBSN is able to show in screen only limited number of points of interest, that is why the more big scale is used the more service points will be refused to be selected by preference filter. • ASSA gets explicitly data about changing possible agents desires from one set of preferences to another one based on changes in scale. ASSA also gets new agent’s view to the services available in the neighbourhood.

42. ASSA Benefits from LBSN (Showing Point) • Showing Point: • LBNS: Input: screen with scaled map, pointed customer location and services on it, customer’s click to certain point of interest; Action: LBSN selects and downloads appropriate online data about requested service, prepares it and delivers to customer’s terminal; Output: screen with online information about point of interest, e.g. recent offerings, prices, contact info, etc. • ASSA gets explicitly focus of the agent’s view, i.e. gets data about agent’s desires, based on type of service the agent is observing now.

43. ASSA Benefits from LBSN (Showing Route) • Showing Route: • LBNS: Input: screen with scaled map, pointed customer location and services on it, customer’s click to the point of interest on the map; Action: LBSN discovers optimal routes from customer’s location to selected point depending on map scale and available transport, prepares appropriate data and delivers it to customer’s terminal; Output: screen with map with highlighted routes between the two points for all available transport facilities. • ASSA gets explicit information about agent’s desire to move towards the selected point and to get appropriate service.

44. ASSA Benefits from LBSN (Call Point) • Call Point: • LBNS: Input: screen with online information about point of interest, customer’s click to “call” button; Action: LBSN via mobile terminal dials telephone number of selected service; Output: Customer is connected to the point of interest. • ASSA gets explicit information about agent’s intentions to get more information about appropriate service and agent’s desire to get this service .

45. ASSA Benefits from LBSN (Order Service) • Order Service: • LBNS: Input: screen with online information about point of interest, customer’s click to “order” button; Action: LBSN via mobile terminal connects customer with appropriate service ordering web page of the selected service; Output: screen with online order form from the selected service web site. • ASSA gets explicit information about agent’s intentions to get selected service.

46. Conclusions (1) • We assumed that location of an agent effects on his beliefs desires and intentions and that the history of observed agent’s mobility can be used to predict his future states; • Formal spatial route language used in this paper is introduced. The assumptions made on the notions belief, desire and intention, and the way of their interactions are discussed and formalised: formal relationships between the intentional notions, and the spatial behaviour of an agent are defined. • The case of using agent architecture for reasoning about the intentions of the customers of a mobile location-based service is also described.

47. Conclusions (2) • The approach introduced opens up a number of possibilities for further work. For example various electronic commerce applications are interested in personalizing their services to the customers by predicting and utilizing customer preferences. For location-aware applications the agent-based analogy of modelling customers’ beliefs, desires and intention in tracked locations might be an important possibility.

48. Acknowledgements Agora Center (University of Jyvaskyla): Agora Center includes a network of good-quality research groups from various disciplines. These groups have numerous international contacts in their own research fields. Agora Center also coordinates and administrates research and development projects that are done in cooperation with different units of university, business life, public sector and other actors. The mutual vision is to develop future's knowledge society from the human point of view. http://www.jyu.fi/agora-center/indexEng.html InBCT Project (2000-2004): Innovations in Business, Communication and Technology http://www.jyu.fi/agora-center/inbct.html